Modelagem matemática de uma usina fotovoltaica de pequeno porte: proposição de modelos intra-horários, direto e reverso, avaliados sob diferentes condições de nebulosidade

• Ano de defesa: 2018
• Autor(a) principal: João Vitor de Navarro Moreira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Geração distribuída; Modelagem matemática; Validação experimental; Energia fotovoltaica; Geração de energia fotovoltaica; Engenharia mecânica
• Link de acesso: http://hdl.handle.net/1843/BUOS-AZ3LDN

The photovoltaic technology is an alternative to generate energy that has a major contribution on the growing share of renewable sources in the overall energy production market. Its development is encouraged in several countries. The Brazilian government develops the technology through public auctions and, recently, regulated the possibility for consumers generate their own energy. On-site generation has exponentially increased its presence in the country, having photovoltaic technology as the source for most of the projects. In this context, this work presents the mathematical modelling of a photovoltaic power plants generation stage throughout two models. The first one, called direct model, is applied to predict the power output from a photovoltaic module given environment parameters, such as solar radiation. The model can be used in real-time monitoring of photovoltaic power plants. The second one, called reverse model, is applied to predict the solar radiation given the photovoltaic modules voltage and current. The model is an alternative way to obtain solar radiation incident where there is no other method to measure it. Both models,proposed for a time resolution greater than hourly, are evaluated for different cloudiness conditions by comparison to data acquired from a small-scale photovoltaic power plant located in Belo Horizonte, Brazil. The results indicate that direct and reverse models are appropriated as prediction methods for power output and solar radiation. The non-systematic error is smaller than 15% on clear sky and cloudy sky. The number is larger only for overcast conditions. The results also indicate opposite tendencies for each models response.The systematic error is positive for direct model and negative for reverse model, revelling that the first one tends to overestimate the response while the response tends to be underestimate by the second one. The two models yet present similar results. The values of systematic error, non-systematic error, maximum and average absolute errors and standard deviation are close figures, indicating that predict solar radiation given the photovoltaic modules voltage and current is just as possible as predict the power output from a photovoltaic module given environment parameters